Skip to main content
HT
HerbDev Application Rescue
← Learning hub

Free guided course · ML 101

Data, features and labels

Turn a useful question into a prediction task.

Beginner · Lesson 1 of 6 · About 8 minutes · No account needed

  • ML 101
  • Practical examples
  • Self-paced learning

Your learning progress

0 of 6 checkpoints complete · Best score: 0 of 18

0 of 6

Accept the cookie notice to save scores and progress on this browser. You can learn and score without accepting; unsaved progress ends when you leave the page.

Answer all three questions correctly to complete a checkpoint. Retry as often as you like; your best score counts. Scores stay on this browser and can be cleared. They do not transfer between devices.

Start the first lesson →

Learn · Try · Check

What you’ll learn

Turn a useful question into a prediction task.

Listen to this lesson

Local AI narration · Quiz answers are kept out of the audio.

Read the audio transcript

ML 101. Data, features and labels. Turn a useful question into a prediction task. Define the prediction first. Machine learning fits a model to patterns in data. Begin with the decision you want to support: predicting a category, estimating a number or finding a useful representation. In supervised learning, examples include a target answer called a label. Classification predicts a category; regression predicts a numerical quantity. Neither task guarantees that a model has discovered the cause of what it predicts. Give each example a clear meaning. Features are the inputs available to the model, such as measurements or pixels. A label is the target associated with an example. For a teaching exercise, imagine photographs labeled “needs review” or “routine.” Specify who assigns those labels and what each category means. Inconsistent labeling teaches inconsistent targets. An annotation is a recorded description or label; it is not automatically objective truth. Think about where the data came from. A dataset made entirely from bright indoor photographs may not represent outdoor field use. Ask which devices, locations and conditions are included, which are missing and whether you have permission to use the material. Keep relevant context such as capture session or site identity so you can later split and evaluate examples sensibly. More examples help only when they are suitable for the task and collected under an appropriate process. HerbDev example. The case study connects capture, review, datasets and evaluation. For this course, the “needs review” classifier is a hypothetical teaching exercise inspired by that workflow; it is not a reported OrganicVision model or performance result. The process: Define a prediction. Then Identify available features. Then Define the target labels. Then Inspect coverage and quality. Try this. For a photo review classifier, identify the input, target and one label ambiguity. Pause the audio to try the exercise. A worked answer and checkpoint questions are available on the lesson page.

Define the prediction first

Machine learning fits a model to patterns in data. Begin with the decision you want to support: predicting a category, estimating a number or finding a useful representation. In supervised learning, examples include a target answer called a label. Classification predicts a category; regression predicts a numerical quantity. Neither task guarantees that a model has discovered the cause of what it predicts.

Give each example a clear meaning

Features are the inputs available to the model, such as measurements or pixels. A label is the target associated with an example. For a teaching exercise, imagine photographs labeled “needs review” or “routine.” Specify who assigns those labels and what each category means. Inconsistent labeling teaches inconsistent targets. An annotation is a recorded description or label; it is not automatically objective truth.

Think about where the data came from

A dataset made entirely from bright indoor photographs may not represent outdoor field use. Ask which devices, locations and conditions are included, which are missing and whether you have permission to use the material. Keep relevant context such as capture session or site identity so you can later split and evaluate examples sensibly. More examples help only when they are suitable for the task and collected under an appropriate process.

The process, step by step
  1. Define a prediction
  2. Identify available features
  3. Define the target labels
  4. Inspect coverage and quality

Try it yourself

For a photo review classifier, identify the input, target and one label ambiguity.

Compare with a worked answer

Input: a photograph. Target: whether it should enter a review queue under a written review policy. Ambiguity: one reviewer flags blur while another does not. Agree on the policy before interpreting a model’s errors.

Check your understanding

Choose one answer for each question, then check your score. Read the explanations and retry any question you missed.

1. In supervised learning, what is a label?

2. Predicting a numerical waiting time is usually which kind of task?

3. What is a problem with inconsistent human labels?

Read the answer guide

1. The target associated with a training example A label supplies the target the model is trained to predict.

2. Regression Regression predicts a numerical quantity; classification predicts a category.

3. They make the target the model should learn unclear The model cannot reliably learn a clearly defined policy if the examples represent conflicting policies.

Was this explanation clear?

Help improve this lesson. With cookies accepted, HerbDev records one current rating per lesson for this browser. Server checks limit repeated submissions; accepted changes may also appear in site analytics. No written answer or account is needed.

You can also share a suggestion with HerbDev.

Review missed questions

Revisit questions missed on your latest checkpoint attempts. These short reviews help you practise; they do not change your best checkpoint scores. Return to the lesson checkpoint when you are ready.

Check a lesson’s answers to build your review list. Saved reviews are available after cookie acceptance.

Keep exploring: What next? · AI vocabulary · AI 101 · Learning hub

Manage browser learning data

“Remove my test data” clears all AI 101 and ML 101 scores, reviews, project checklists and rating selections, plus saved vocabulary and recent history in this browser. Test entries cannot be separated from other learning progress. Cookie preferences, security protections, submitted feedback, contact emails and past analytics are kept.